2.6 Statistical analysis
Descriptive data are reported as the median with interquartile ranges for continuous data and numbers with percentages for categorical data. To assess the association between baloxavir use and those outcomes, we used univariate and multivariable logistic regression analyses adjusting for demographic data (age and sex) and comorbid conditions (five categories).
We conducted a prespecified subgroup analysis. Baloxavir was compared with each NAI. This analysis was planned since laninamivir is thus far licensed exclusively in Japan [17]. For this analysis, only risk-adjusted odds ratio (aOR) was reported. We also planned a subgroup analysis stratified by virus type, but influenza type B infection was only infrequently found in our data set (<1%); consequently, we did not perform this subgroup analysis.
A P value <0.05 was considered statistically significant. P value adjustment for multiple comparisons was not conducted, and thus the findings for secondary outcomes and subgroup analysis should be viewed as exploratory, hypothesis-generating testing [18]. All analyses were done with R statistical environment (https://cran.r-project.org/), under version 3.61.